Differential Stability of Biogenesis Intermediates Reveals aCommon Pathway for Aquaporin-1 Topological Maturation*

Received for publication, August 30, 2004, and in revised form, October 27, 2004Published, JBC Papers in Press, October 28, 2004, DOI 10.1074/jbc.M409920200

Teresa M. Buck and William R. Skach‡

From the Molecular Medicine Division, Department of Medicine, Oregon Health Sciences University,Portland, Oregon 97201

Topological studies of multi-spanning membrane pro-teins commonly use sequentially truncated proteinsfused to a C-terminal translocation reporter to deducetransmembrane (TM) segment orientation and key bio-genesis events. Because these truncated proteins repre-sent an incomplete stage of synthesis, they transientlypopulate intermediate folding states that may or maynot reflect topology of the mature protein. For example,in Xenopus oocytes, the aquaporin-1 (AQP1) water chan-nel is cotranslationally directed into a four membrane-spanning intermediate, which matures into the six mem-brane-spanning topology at a late stage of synthesis(Skach, W. R., Shi, L. B., Calayag, M. C., Frigeri, A., Lin-gappa, V. R., and Verkman, A. S. (1994) J. Cell Biol. 125,803–815 and Lu, Y., Turnbull, I. R., Bragin, A., Carveth,K., Verkman, A. S., and Skach, W. R. (2000) Mol. Biol. Cell11, 2973–2985). The hallmark of this process is that TM3initially acquires an Nexo/Ccyto (Type I) topology andmust rotate 180° to acquire its mature orientation. Incontrast, recent studies in HEK-293 cells have sug-gested that TM3 acquires its mature topology cotrans-lationally without the need for reorientation (Dohke,Y., and Turner, R. J. (2002) J. Biol. Chem. 277, 15215–15219). Here we re-examine AQP1 biogenesis and showthat irrespective of the reporter or fusion site used,oocytes and mammalian cells yielded similar topologicresults. AQP1 intermediates containing the first threeTM segments generated two distinct cohorts ofpolypeptides in which TM3 spanned the ER membranein either an Ncyto/Cexo (mature) or Nexo/Ccyto (imma-ture) topology. Pulse-chase analyses revealed that theimmature form was predominant immediately aftersynthesis but that it was rapidly degraded via the pro-teasome-mediated endoplasmic reticulum associateddegradation (ERAD) pathway with a half-life of lessthan 25 min in HEK cells. As a result, the mature to-pology predominated at later time points. We concludethat (i) differential stability of biogenesis intermedi-ates is an important factor for in vivo topological anal-ysis of truncated chimeric proteins and (ii) cotransla-tional events of AQP1 biogenesis reflect a commonAQP1 folding pathway in diverse expression systems.

Polytopic membrane proteins are synthesized and orientedin the endoplasmic reticulum (ER)1 by the ribosome and Sec61translocon complex (1–3). In the simplest model, topology ofeach transmembrane (TM) segment is established in a vectoraland sequential manner (N to C termini) (4) as independentsignal anchor and stop transfer sequences alternately gate thetranslocon and the ribosome-translocon junction and direct TMsegment integration into the lipid bilayer (cotranslationalmodel) (5–7). However, a growing body of evidence has demon-strated that the final topology of many native proteins is notnecessarily established cotranslationally but rather throughcooperative interactions between topogenic determinants (TMsegments) located within different regions of the polypeptide(post-translational model) (8–16).

One example of the post-translational model occurs duringthe biogenesis of aquaporin-1 (AQP1), a hydrophobic mem-brane protein of �29 kDa that exists as a homo-tetramer in cellmembranes. AQP1 is a member of the MIP (major intrinsicprotein) family (17, 18). In its mature form it exhibits a char-acteristic topology with six TM segments and two additionalshort helical regions flanked by conserved NPA motifs that foldinward within the plane of the membrane to form a monomeric,water-selective pore (19, 20). AQP1 is expressed in diverse celltypes and is localized in the kidney to the proximal tubule anddescending limb of the loop of Henle where it plays a major rolein renal water reabsorption (18, 21).

Early biogenesis studies of AQP1 in cell-free systems andmicroinjected Xenopus oocytes revealed a novel mechanism inwhich only four of its six transmembrane segments cotransla-tionally acquired a membrane spanning topology (22). Thisfour-spanning intermediate later matures in the ER membraneto form the final six-spanning structure (23, 24). AQP1 biogen-esis differs from the cotranslational pathway utilized by a closehomolog, AQP4 (25), in part because hydrophilic residueswithin the N terminus of TM2 (Asn49 and Lys51) disrupt stoptransfer sequence and allow TM2 to transiently pass throughthe translocon and into the ER lumen (24). As a result, whenTM3 emerges from the ribosome, it acts as a stop transfersequence to terminate translocation and cotranslationallyadopt an Nexo/Ccyto (Type I) topology. A four-spanning inter-mediate is synthesized because TM4 does not reinitiate trans-location, and TM5 and TM6 act as signal and stop transfersequences, respectively (see Fig. 1A). The defining feature ofthis folding intermediate is that TM3 is initially oriented withits C terminus positioned in the cytosol. In Xenopus oocytes,TM3 topology is “corrected,” because it rotates 180° about theplane of the membrane to acquire its mature topology during a

* This work was supported by National Institutes of Health GrantsGM53457 and DK51818, an American Heart Association EstablishedInvestigator Grant (to W. R. S.), and National Institutes of HealthGrant HL007781 (to T. M. B.). The costs of publication of this articlewere defrayed in part by the payment of page charges. This article musttherefore be hereby marked “advertisement” in accordance with 18U.S.C. Section 1734 solely to indicate this fact.

‡ To whom correspondence should be addressed: Division of Molecu-lar Medicine, Oregon Health Sciences University, 3181 SW Sam Jack-son Park Rd., Portland, OR 97201. Tel.: 503-494-7322; Fax: 503-494-7368; E-mail: skachw@ohsu.edu.

1 The abbreviations used are: ER, endoplasmic reticulum; AQP, aqua-porin; TM, transmembrane; spanning, membrane-spanning; EGFP, en-hanced green fluorescent protein; ERAD, endoplasmic reticulum asso-ciated degradation; PNGase F, N-glycosidase F.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 280, No. 1, Issue of January 7, pp. 261–269, 2005© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org 261

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

later stage or following the completion of AQP1 synthesis (23).Recent work has suggested that the AQP1 biogenesis path-

way may be dependent on cell type (26). A study by theTurner group (26) examined the topology of truncated con-structs in which variable numbers of AQP1 TM segmentswere placed into a chimeric protein containing N- and C-terminal reporter domains derived from EGFP and the �-sub-unit of the H/K-ATPase, respectively. The topology of TMsegments was inferred by N-linked glycosylation of the C-terminal reporter in intact HEK-293 cells. In contrast toresults in oocytes, the Turner group found that TM3 hadalready acquired its mature Ncyt/Cexo topology (e.g. the re-porter was glycosylated) in constructs containing only thefirst three AQP1 TM segments and concluded that AQP1biogenesis therefore occurs in a cotranslational fashion with-out reorientation from a four-spanning intermediate.

These studies have lead to several unanswered questionsregarding AQP1 biogenesis. Different reporter domains (pro-lactin versus H/K-ATPase �-subunit derivatives), translocationassays (protease protection versus glycosylation) and fusionsites could potentially account for the different apparent topol-ogy observed for TM3. Because C-terminal reporters are rou-tinely used to study polytopic protein biogenesis, these factorsare of more than just academic interest. In addition, althoughtruncated proteins provide important structural information ata relatively defined point of synthesis, they lack C-terminalsequence information and therefore must be viewed as popu-lating intermediate folding states. As such they are potentialcandidates for recognition by ER quality control machinery.Consistent with this we had previously observed that certainAQP1 fusion proteins are relatively unstable in oocytes (27). Afinal possibility proposed by the Turner group is that oocytesand mammalian cells handle topological information in funda-mentally different ways. If true, this would have major impli-cations for protein biogenesis.

Xenopus oocytes efficiently express diverse aquaporins andhave been extensively used to study AQP biosynthesis, traffick-ing, and function (28–32). However, a direct comparison ofbiosynthetic mechanisms in oocyte and mammalian cells hasnot previously been carried out. We therefore systematicallyexamined AQP1 fusion proteins containing two, three, or fourTM segments in both Xenopus oocytes and HEK-293 cells todetermine the origin of previous discrepancies. We now showthat irrespective of the reporter or fusion site examined, bothsystems yielded similar topologic results. AQP1 intermediatescontaining the first three TM segments gave rise to two distinctcohorts of polypeptides in which TM3 spanned the ER mem-brane in either an Ncyto/Cexo (mature) or Nexo/Ccyto (imma-ture) topology. Careful pulse-chase analyses revealed that theimmature form predominated immediately after synthesis inboth cell types but that it was rapidly degraded via the protea-some-mediated ERAD pathway. As a result, the mature topol-ogy predominated 1–2 h after synthesis in HEK cells. Weconclude that differential stability of biogenesis intermediatesis an important factor for in vivo analysis of truncated chimericproteins and that cotranslational events of AQP1 biogenesisare conserved in diverse expression systems.

MATERIALS AND METHODS

cDNA Construction—Plasmids AQP1P77.P, AQP1L139.P, andAQP1P169.P are described previously as clones 3, 6, and 7 (22);AQP1S66.P, AQP1T120.P, and AQP1L164.P were generated using thesame strategy by PCR amplification of AQP coding region using anti-sense oligonucleotides: S66, GCTGATGGTCACCCCACTCTGCGCCA-GCGTGGC; T120, AAGCGAGGTCACCGTCAGGGAGGAGGTGAT;and L164, GGGGGCGGTCACCCCAAGGTCACGGCGCCTCCG. Theresulting constructs contain AQP1 residues 1–66, 1–77, 1–120, 1–139,1–164, or 1–169 followed by the C-terminal 142 amino acids of bovine

prolactin, a passive translocation reporter (5). Plasmid pEGFP.AQP1T120.� was kindly provided by R. James Turner and is derivedfrom pEGFP-C3 (Clontech, Palo Alto, CA) (26). It contains AQP1 codons1–120 fused between EGFP and the C-terminal 177 amino acids of theH/K-ATPase �-subunit. pEGFP.AQP1S66.�, pEGFP.AQP1P77.�,pEGFP.AQP1L139.�, pEGFP.AQP1L164.�, and pEGFP.AQP1P169.�were generated by PCR amplification of AQP1 cDNA using a senseoligonucleotide (TGAGTAGATCTCATGGCCAGCGAGTTCAAG) andcorresponding antisense oligonucleotides S66 (TGAGTAAGCTTCCAC-TCTGCGCCAGCGTC), P77C (AGTGTGAAGCTTCCCGGGTTGAGGT-GGGCCCG), L139 (ATCTCGAAGCTTCCCAGGCCCTGGCCCGAGT-TC), L164 (TGAGTAAGCTTCCAAGGTCACGGCGCCT), and P169 (C-CGATGAAGCTTCCGGGGGCTGAGCCACCAAGG) encoding BglII(sense) and HindIII (antisense) restrictions sites that were used toligate DNA fragments into the EGFP-� subunit cassette. The constructsencode AQP1 residues 1–66, 1–77, 1–139, 1–164, or 1–169 flanked byEGFP and �-subunit. The full-length AQP1 construct was generated byligating AQP1 cDNA into the mammalian expression vector pEGFP-N3(Clontech) between the HindIII and NotI restriction sites resulting in afull-length AQP1 construct without an EGFP tag. The sequence of allPCR-amplified DNA was verified by automated DNA sequencing.

Xenopus laevis Expression—mRNA was transcribed in vitro with SP6RNA polymerase (New England Biolabs, Beverly, MA) using 2 �g ofplasmid DNA in a 10-�l volume at 40 °C for 1 h as previously described(22). The aliquots were used immediately or frozen in liquid nitrogen andstored at �80 °C. 2 �l of transcript was mixed with 50 �Ci of [35S]methi-onine (0.5 �l of a 10� concentrated Tran35S-label; ICN Pharmaceuticals,Irvine, CA), and 50 nl was injected into stage VI X. laevis oocytes (50nl/oocyte) on an ice-cold stage. The oocytes were incubated at 18 °C inMBSH (88 mM NaCl, 1 mM KCl, 24 mM NaHCO3, 0.82 mM MgSO4, 0.33mM Ca(NO3)2, 0.41 mM CaCl2, 10 mM HEPES, pH 7.4, 50 �g/ml genta-micin, 100 units/ml penicillin, and 100 �g/ml streptomycin sulfate).

Protease Protection and Pulse-Chase Assays in Oocytes—The oocyteswere injected as described above, incubated at 18 °C for 3 h, andhomogenized by hand in 3 volumes of 0.25 M sucrose, 50 mM KAc, 5 mM

MgAc2, 1.0 mM dithiothreitol, 50 mM Tris-Cl, pH 7.5. The homogenateswere divided into three 10-�l aliquots on ice. Proteinase K (RocheApplied Science) was added (final concentration, 0.2 mg/ml) in thepresence or absence of 1% Triton X-100. The reactions were incubatedon ice for 1 h and rapidly mixed with phenylmethylsulfonyl fluoride (10mM) and boiled in 10 volumes of 1% SDS, 0.1 M Tris-HCl, pH 8.0, for 5min. The samples were then diluted in 10 volumes of buffer A (100 mM

Tris, pH 8.0, 100 mM NaCl, 1% Triton X-100, 2 mM EDTA), incubated at4 °C for 15 min, and centrifuged at 16,000 � g for 10 min at 4 °C toremove insoluble debris. Efficiency of the assay was regularly assessedusing a known secretory control protein and was �80%. For pulse-chaseassays oocytes were injected as described and incubated at 18 °C for 2 h,and the medium was then replaced with fresh MBSH containing 2 mM

methionine. The oocytes were harvested at the specified time points andprocessed as above.

HEK-293 Pulse-Chase Studies—HEK-293 cells were cultured in Dul-becco’s modified essential medium (Fisher) supplemented with 100units/ml penicillin, 100 �g/ml streptomycin sulfate, and 10% heat-inactivated fetal bovine serum (Invitrogen). The cells were grown onplastic in a humidified incubator at 37 °C and 5% CO2 and were pas-saged every 3–4 days. The cells were transfected at 50–60% confluencewith TransIT-LT transfection reagent (Mirus, Madison, WI) accordingto the manufacturer’s instructions (3 �g of cDNA and 12 �l of Tran-sIT-LT reagent/60-mm plate). Transfection efficiencies were �40% asdetermined by EGFP fluorescence. 48 h after transfection media wasreplaced with 1 ml of methionine-cysteine-free medium for 30 min. Thecells were then pulse-labeled with 80 �Ci of [35S]methionine for 15 min,and the medium was removed and replaced with fresh complete me-dium for the indicated duration of the chase. The cells were lysed on icewith 1 ml of radioimmune precipitation assay buffer (0.1% SDS, 1%Triton X-100, 1% dexoycholate, 150 mM NaCl, 10 mM Tris-Cl, pH 8.0, 2.5mM MgCl2, 1� protease inhibitors III (CalBiochem, San Diego, CA), and1 mM phenylmethylsulfonyl fluoride) for 30 min on ice and passedthrough a 26-gauge needle three times. The samples were then clarifiedat 16,000 � g for 12 min.

Immunoprecipitation—Clarified oocyte or HEK cell homogenateswere incubated with anti-prolactin antisera (ICN Biomedicals, CostaMesa, CA) at 1:2000 dilution, polyclonal anti-EGFP antibody (Molecu-lar Probes, Eugene, OR) at 1:2000 dilution, or polyclonal rabbit anti-AQP1 antisera raised against purified AQP1 protein (generously pro-vided by A. Verkman, UCSF, San Francisco, CA) at 1:1000 dilution and5.0 �l of protein-A Affigel (Bio-Rad). The samples were rotated for 10 hat 4 °C (oocyte homogenates) and 4 h (HEK cells) prior to washing three

Stability of AQP1 Biogenesis Intermediates262

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

times with solubilization buffer and twice with 100 mM NaCl, 100 mM

Tris-Cl, pH 8.0. The samples were boiled in protein SDS loading buffer(33) and analyzed by SDS-PAGE, EN3HANCE (PerkinElmer Life Sci-ences) fluorography. The intensity of recovered bands was quantitatedusing a Bio-Rad personal Molecular PhosphorImager Fx (Kodakscreens, Quantity-1 software).

PNGase F Digests—Beads from immunoprecipitations were resus-pended after the final wash in 15 �l of 0.1 M Tris-Cl, pH 7.5, and 0.3 �lof PNGase F (New England Biolabs, Beverly, MA), and incubated at37 °C for 3 h. The samples were then analyzed by SDS-PAGE asdescribed above.

RESULTS

C-terminal translocation reporters placed after connectingpeptide loops have been widely used to study membrane pro-tein topology and biogenesis (6, 12, 13, 22, 34–39). The ration-ale for this approach is that topology of TM segments and theirconnecting loops is controlled by the action of topogenic deter-minants encoded within the nascent polypeptide. TM segmentsthat function as signal (anchor) sequences open the Sec61translocon channel and initiate translocation of polypeptideinto the ER lumen. All else being equal, translocation willcontinue until a second TM segment is synthesized that func-tions as a stop transfer sequence to close the translocon, relaxthe ribosome-translocon junction, and allow the next peptideloop to enter the cytosol. In this manner, topology is deter-mined cotranslationally as the nascent polypeptide emergesfrom the ribosome. A passive reporter domain (i.e. one contain-ing no intrinsic topogenic information) fused to a peptide loopdownstream of a signal sequence will therefore follow the loopinto the ER lumen, whereas a reporter located downstream ofa stop transfer will, in turn, remain in the cytosol.

Two variations on this strategy have been employed to mapthe cotranslational topology of AQP1. In the first approach, aC-terminal reporter domain derived from the secretory proteinprolactin was fused to each AQP1 peptide loop, and topologywas determined in vitro and in microinjected Xenopus oocytesusing a protease protection assay (22). The second approachused a different chimeric cassette containing an N-terminalEGFP domain and a C-terminal reporter derived from the�-subunit of the H/K-ATPase. Topology of the �-subunit wasthen determined in HEK-293 cells based on N-linked glycosyl-ation, which occurs exclusively in the ER lumen (26, 40, 41).Key differences between these studies are summarized in Fig.1. In oocytes, reporters fused to the TM2–3 loop (residue Pro77)or the TM3–4 loop (Leu139) were located in the ER lumen andcytosol, respectively (Fig. 1A) (22). Surprisingly this was dif-ferent from their expected topology in the mature protein. Twosubsequent studies confirmed these results and demonstratedthat AQP1 TM segments 2–4 cotranslationally acquire an im-mature topology that is subsequently converted to the maturetopology by an internal 180° rotation of TM3 (23, 24). In contrast,glycosylation patterns of the �-subunit domain lead to the con-clusion that the TM2–3 and TM3–4 loops acquire their propertopology cotranslationally in the ER of mammalian cells andtherefore do not undergo a topological reorientation (Fig. 1B).

Although similar in many ways, these studies differed inseveral important aspects including the location of fusion sites,the reporter domain used, and the cell expression system. Wetherefore undertook a systematic comparison of AQP1 biogen-esis in oocytes and mammalian cells using similar truncationsites and reporters to resolve these discrepancies. Because themajor differences involved the initial orientation of TM3 (TypeI in oocytes and Type II in HEK cells), we focused our attentionprimarily on the topology of TM3 and its immediate flankingresidues. Topology of other regions such as the N and C terminiand TM5–6 are well established and were therefore not re-tested here (Fig. 1).

Reporter Fusion Site Does Not Affect Outcome of Early Bio-genesis Experiments in Oocytes—We first addressed whethertopological differences might arise from the use of differentfusion sites. In particular, up to 10–15 flanking residues caninfluence topogenic activities of TM segments (42). Thus fusionsites very close to the TM C terminus such as those used in theEGFP-�-subunit chimeras might delete potentially importanttopogenic information. Alternatively, the fusion site might in-troduce new flanking residues (from the �-subunit) that couldinadvertently alter TM2 behavior. AQP1 fusion sites weretherefore tested head-to-head by placing the prolactin reporterat both truncations sites downstream of TM2 (Ser66 and Pro77),TM3 (Thr120 and Leu139), and TM4 (Leu164 and Leu169 (Fig.2A). Topology was determined by protease protection as de-scribed under “Materials and Methods.” Consistent with ourprevious results, fusion to the TM2–3 peptide loop at eithertruncation Ser66 or Pro77 resulted in similar translocation ofthe reporter (80–83%; Fig. 2B, lanes 1–6). Similarly, the �-sub-unit-TM2 fusion site used by Dohke et al. (46) had only amodest effect on translocation of the P-reporter, decreasingtranslocation by 5–10% (data not shown). In contrast, when thereporter was placed in the TM3–4 loop (truncations Thr120 andLeu139) or the TM4–5 loop (truncations Leu164 and Leu169), itwas �85–90% and �95% cytosolic, respectively (Fig. 2B, lanes7–18). As was observed previously, �50% of fusion proteinstruncated after TM3 (TM3.P and TM4.P) underwent N-linkedglycosylation at residue Asn42, which resulted in a 3-kDa shiftin migration (22).

These results thus support previous topological findings inXenopus oocytes and demonstrate that different topologicalorientations obtained for TM2 and TM3 (shown in Fig. 1) werenot likely caused by subtle modification of downstream flank-ing residues due to different fusion sites. In all cases, TM2 didnot function as an efficient stop transfer sequence to terminatetranslocation, and TM3 initially spanned the membrane withits C-terminal flanking residues oriented toward the cytosol.

Truncated AQP1 Constructs Generate Multiple BiogenesisIntermediates with Differential Stability in Oocytes—One limi-

FIG. 1. Two models for AQP1 biogenesis. Topology of reporterfusion sites used for experiments in Xenopus oocytes (A) and HEK-293cells (B) are indicated. Fusion sites in which topology differs betweenthese systems are shown in black, whereas fusion sites with the sametopology are shown in gray. In oocytes AQP1 is initially synthesized asa four-spanning intermediate that is converted into the mature six-spanning form. In HEK-293 cells the reporter did not identify theimmature topology (26).

Stability of AQP1 Biogenesis Intermediates 263

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

tation to all topology mapping studies is that results are oftennot absolute. For example, variable degrees of glycosylationwere observed for several EGFP-AQP-�-subunit chimeras inHEK cells, particularly for fusion sites in the TM2–3 andTM3–4 loops (26). Similarly, in our original analysis we used acut-off of �80% protease protection to define whether a peptideloop segment was translocated into the ER lumen (22). Asshown in Fig. 2B, the majority of the reporter fused to residuesThr120 or Leu139 was accessible to cytosolic protease. However,a small fraction (10–15%) appeared to be resistant to digestion.Although this could potentially be caused by subtle variationsin reporter folding and efficiency of proteolysis, this minorfraction was completely degraded in the presence of nondena-turing detergent (Fig. 2B, lanes 9 and 12). This raised thepossibility that nascent polypeptides with protected versus ac-cessible reporters might represent separate cohorts of proteinswith different topological structures.

We therefore used pulse-chase metabolic labeling to test thetopology of AQP1 fusion proteins truncated after TM3 at resi-dues Thr120 and Leu139. Oocytes were simultaneously injectedwith mRNA and [35S]methionine to initiate rapid radiolabelingof newly synthesized protein. Labeling gradually tapers offafter several hours because of dilution by medium and oocytemethionine stores (Ref. 43 and data not shown). In some casesthe chase was performed in the presence of excess unlabeledmethionine, but this had little effect given the long time courseused (data not shown). Incubation times were chosen based onprevious studies demonstrating that processing and transportproteins out of the oocyte ER is significantly slower than mam-malian cells and occurs over the time frame of many hours (43,44). At the end of the incubation, reporter topology of remain-ing polypeptides was determined by protease protection. Twoimportant results were obtained from these experiments. First,both constructs, regardless of the fusion site, were quite unsta-ble and underwent significant degradation during the chaseperiod (Fig 3A, compare lanes 1, 4, and 7 with lanes 10, 13, and16). After 24 h the amount of protein remaining had decreasedby 60–80% (Fig. 3B). In contrast, there was little or no de-

crease in the amount of protein that yielded a protease pro-tected reporter. This resulted in a 3–4-fold increase in thefraction of remaining chains in which TM3 C-terminal residuesappeared to reside in the ER lumen (Fig. 3, C and D).

Although not conclusive in themselves, these results sug-gested a scenario in which synthesis of the first three AQP1 TMsegments provides insufficient topogenic information to effi-ciently direct TM segments into their proper topology. Mostpolypeptides, (�85% in oocytes) can only achieve a two-span-ning topology in which TM2 does not yet stably span the mem-brane and in which TM3 is oriented in an Nexo/Ccyto topology.However, in a relatively small fraction of polypeptides, TM3does acquire its proper Ncyt/Cexo topology, which is reflected bytranslocation of the TM3–4 loop and its attached reporter (Fig.3C). Thus two different isoforms appear to be generated inoocytes when synthesis of TM3 and TM4 is completed, andsynthesis of TM5–6 has not yet begun. Whereas the two-span-ning topology initially predominates, selective degradation ofthis isoform leads to a relative increase in the more stablethree-spanning isoform over time.

AQP1 Fusion Proteins Generate Multiple Topological Iso-forms in Mammalian Cells—If truncated AQP1 constructsgave rise to multiple topological isoforms in mammalian cells,then differential stability might explain some of the topologicaldiscrepancies demonstrated in Fig. 1. We therefore re-evalu-ated chimeric fusion proteins previously used by the Turnergroup to better define AQP1 biogenesis in mammalian cells.AQP1 fragments were inserted into the EGFP-� subunit chi-mera described previously at two different truncation sitesafter TM2, TM3, and TM4 (26). As shown in Fig. 4A, the�-subunit contains five potential N-linked glycosylation sitesthat provide a read-out for translocation into the ER lumen.Constructs were transiently expressed in HEK-293 cells andpulse-labeled with [35S]methionine for 2 h, and EGFP reactiveproteins were immunoprecipitated and analyzed by SDS-PAGE before and after removal of N-linked sugars by PNGaseF. When AQP1 was truncated at residues Ser66 and Pro77, 64and 57% of EGFP-reactive protein was present in a glycosy-

FIG. 2. Location of fusion sites doesnot affect early AQP1 biogenesisevents in oocytes. A, diagram of AQP1fusion proteins showing TM segments(black rectangles) and prolactin-derived(P) reporter (gray rectangles). B, con-structs in A were expressed in oocytes,and reporter topology was determined byproteinase K (PK) digestion in the pres-ence or absence of Triton X-100 (Tx-100).Shown is a representative autoradiogramof prolactin reactive immunoprecipitatedproducts. The percentage of reporter pro-tected from proteinase K digestion wasnormalized to a control secretory protein.Topology of fusion sites are illustrated un-der autoradiogram. The data show aver-age of three or more experiments �S.E.M.

Stability of AQP1 Biogenesis Intermediates264

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

lated form (Fig. 4B, lanes 1–4). This is consistent with previousreports that found �50% glycosylation in the identical Ser66

construct at steady state. However, in contrast to the previousconclusions that TM2 spanned the membrane cotranslation-ally, we would argue that TM2 does not efficiently terminatetranslocation in HEK cells. Rather its C-terminal residues passinto the ER lumen in a major fraction of nascent polypeptides.

The �-subunit reporter was also predominantly glycosylatedwhen fused to residues Thr120 or Leu139. Most recovered pro-tein (75–82%) migrated at a size of �62–65 kDa, whereas onlyfaint bands were observed at �50–53 kDa (Fig. 4B, lanes 5–8).PNGase F digestion confirmed that the larger bands repre-sented glycosylated forms of the 50-kDa chimeric protein.These results are consistent with steady state immunoblots(26) of the Thr120 truncation in which the �-subunit reporterwas glycosylated in �80% of polypeptides. The simplest expla-nation for these results is that TM3 is indeed predominantlyoriented in its mature Ncyt/Cexo topology. However, a smallproportion of nonglycosylated protein was also isolated (Fig.4B, upward arrows). Although it is possible that this mightreflect inefficient glycosylation of the reporter (45), this seemsunlikely because of the large number of consensus sites avail-able in the �-subunit. Indeed, lack of glycosylation of thisreporter has been used as a direct readout for its cytosoliclocation (26, 38, 46). Polypeptides with nonglycosylated report-ers therefore more likely represent a minor cohort of AQP1 thatexhibits an alternate topology in which the reporter remains inthe cytosol. Thus, like oocytes, mammalian cells appear to

generate two different isoforms from constructs containing thefirst three AQP1 TM segments, but the relative proportion oftwo- and three-spanning isoforms would appear to be reversed(Fig. 4C).

Differential Stability of AQP1 Topological Intermediates inHEK-293 Cells—Based on our observations in oocytes, we fur-ther investigated the relationship between glycosylated andunglycosylated (three-spanning versus two-spanning) AQP1 fu-sion proteins by pulse-chase labeling in HEK cells. Because ofrapid processing, very short pulse times (15 min) were used tocapture AQP1 chimeras as close to the completion of synthesisas possible. These experiments revealed that immediately aftersynthesis, AQP1 chimeras containing the first three TM seg-ments (truncations Thr120 or Leu139) were glycosylated in lessthan 50% of total protein (Fig. 5). Glycosylated proteins wereinitially present as a series of bands between 53 and 65 kDa insize that became fully glycosylated within 1 h. Remarkably,50–60% of both chimeras were initially present as nonglycosy-lated protein that rapidly disappeared with a half-life of �25min. Thus within 1 h of synthesis, �80% of remaining proteincontained a glycosylated reporter.

Although it is possible that the reporter might have under-gone a delayed glycosylation after reaching the ER lumen, wedo not believe this is the case because the total amount ofglycosylated protein did not change significantly during thechase period when all glycosylated bands were quantitated(data not shown; see also Fig. 6). Rather our data indicate thata significant fraction of these chimeras is synthesized in a

FIG. 3. Differential stability of bio-genesis isoforms truncated followingTM3. A, autoradiogram of AQP1 trun-cated in the TM3–4 loop and fused to theprolactin-derived reporter at residuesThr120 and Leu139 as indicated. The con-structs were expressed in oocytes for thetimes indicated and homogenates were di-gested with proteinase K in the presenceor absence of Triton X-100 (TX-100). B,results from A were quantitated, and thepercentage of reporter protected at eachtime point was plotted. The results showthe averaged values from three independ-ent experiments � S.E.M. C, model of iso-form profile. At early points the two-span-ning isoform predominates, whereas theproportion of three-spanning isoform in-creases �3-fold over the time course ofthe experiment.

Stability of AQP1 Biogenesis Intermediates 265

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

two-spanning topology where TM3 is cotranslationally orientedin an Nexo/Ccyt topology. Just as in oocytes, this topologicalisoform is very unstable and selectively degraded, leaving theminor isoform to predominate at later time points. Because ofits very short half-life, it is likely that some of the two-spanningisoform is degraded during the pulse period and that the pro-tein present at time � 0 represents an underestimate of theactual fraction of the two-spanning isoform (Fig. 6). These dataare therefore very similar to the results obtained in oocytes anddemonstrate that regardless of the reporter, oocytes and mam-malian cells both seem to handle AQP1 cotranslational biogen-esis in a similar fashion.

Selective Degradation of AQP1 Biogenesis Intermediates Oc-curs via the Proteasome-mediated ERAD Pathway—Misfoldedproteins in the ER membrane are usually recognized by qualitycontrol machinery and degraded by the ubiquitin-proteasome(ERAD) pathway. We therefore tested whether selective deg-radation of the two-spanning isoform of AQP1 occurred via theproteasome. Pulse-chase studies performed in the presence of

the proteasome inhibitor MG132 revealed that nonglycosylated(two-spanning) isoforms were dramatically stabilized upon pro-teasome inhibition (Fig. 6, A and C). Similar results wereobserved using another proteasome inhibitor N-acetyl-leucyl-leucyl-norleucinal (data not shown). The relative fraction of thetwo-spanning isoforms was increased to �70% at initial timepoints, likely as a result of blocking degradation that took placeeven during the pulse period as mentioned above. In addition,the ratio of two- and three-spanning isoforms remained nearlyconstant in the presence of MG132 because selective degrada-tion of the two-spanning isoform was reduced (Fig. 6, B and D).Importantly, no conversion of the nonglycosylated protein tothe glycosylated form was observed. Therefore the predomi-nance of the three-spanning topology observed at late timepoints does not result from delayed glycosylation of a translo-cated �-subunit reporter. We conclude that the changes ob-served in the topological profiles of AQP1 chimeras containingthe first three TM segments result from selective degradationof the two-spanning, relative to the three-spanning, isoform.

Full-length AQP1 Is Stable in HEK-293 Cells—We previouslyshowed that in Xenopus oocytes, completion of AQP1 synthesisprovides additional folding information that converts TM3 fromits immature to its mature conformation (23). Thus in the full-length protein, the six-spanning topology is not a result of selec-tive degradation but is derived from a rearrangement of TMs2–4. However, it remained theoretically possible that in mam-malian cells, full-length AQP1 could be synthesized as a four-spanning intermediate that was rapidly degraded and that thiscould result in the predominant mature six-spanning topology.Because the majority of AQP1 is initially synthesized with TM3in reverse orientation (Fig. 6), we would expect �50% of theprotein to be rapidly degraded by the proteasome-ERAD path-way. To test this hypothesis, the stability of full-length AQP1 was

FIG. 4. 2-h labeling of HEK-293 cells shows results consistentwith previous studies. A, diagram of chimeric proteins transfectedinto HEK-293 cells. The open circles represent N-linked glycosylationsites on AQP1 and �-subunit. EGFP and �-subunit domains and AQP1TM segments are indicated. B, autoradiogram of 2-h pulse-labeledHEK-293 cells immunoprecipitated with GFP antisera prior to andafter PNGase F digestion. The downward arrows indicate glycosylatedproteins as documented by PNGase F digestion. The upward arrowsindicate nonglycosylated protein. The percentage of glycosylation ofconstructs is indicated under autoradiogram. C, diagram of the topo-logical isoforms deduced from the Thr120 and Leu139 chimeric proteins.

FIG. 5. Differential stability of AQP1 isoforms in HEK-293cells. A, cells were transfected with constructs (Fig. 4), labeled with[35S]methionine for 15 min, chased with complete media for the indi-cated times, lysed, and subjected to SDS-PAGE and autoradiography.Nonglycosylated (�g) and glycosylated (�g) proteins are indicated. B,at time � 0, �50% of protein was recovered as a 50-kDa nonglycosyl-ated band that rapidly disappeared during the chase period. At latertime points the three-spanning (glycosylated) protein predominated.Half-lives of the two-spanning (nonglycosylated) and three-spanning(glycosylated) isoforms were �25 min and �120 min, respectively. Thedata show the averages of three or more experiments � S.E.M.

Stability of AQP1 Biogenesis Intermediates266

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

examined by pulse-chase analysis in HEK cells. The results inFig. 7 clearly demonstrate that in contrast to truncated con-structs, full-length AQP1 was very stable throughout the chaseperiod. Moreover, no significant differences in expression level orstability were observed in the presence of MG132 under ourconditions during the time course expected for topological matu-ration in the ER. Given the very short pulse labeling period usedin our study, it is extremely unlikely that degradation of theunstable four-spanning isoform could account for these findings.Thus we conclude that protein initially synthesized with TM3 inthe reverse orientation is converted into a more stable structureas additional folding information is provided by synthesis of Cterminus residues.

DISCUSSION

Previous studies proposed two different pathways by whichAQP1 acquires its mature six-spanning topology in cells. InXenopus oocytes, biogenesis proceeds from a four-spanning in-termediate into a mature six-spanning topology during latestages of synthesis (22–24). The hallmark of this unusual fold-ing pathway is that TM3 is initially oriented in an Nexo/Ccyt

topology and must rotate 180° to acquire its proper topology

and position TM2 and TM4 within the membrane. In contrast,a recent study failed to observe this intermediate TM3 orien-tation and concluded that mammalian cells utilize a fundamen-tally different biogenesis mechanism in which the matureAQP1 topology is established directly by cotranslational events(26). The current study now reconciles these differences bydirectly comparing early AQP1 biogenesis events in oocytesand mammalian cells using systematically truncated fusionproteins and two different C-terminal reporter domains.

Our results demonstrate a high degree of conservation in theunique features of AQP1 topogenesis that was independent ofthe fusion site, reporter domain, topological assay, or expres-sion system utilized. Specifically, as the second TM segmentemerges from the ribosome, it failed to efficiently terminatetranslocation of its C terminus flanking residues in at least50% of nascent polypeptides in both Xenopus oocytes and mam-malian cells. As a result, TM3 subsequently functions as a stoptransfer sequence to adopt an Nexo/Ccyt topology in which its Cterminus is transiently oriented in the cytosol. This sequence ofevents gives rise to an intermediate topological state in whichonly two of the first three AQP1 TM segments synthesized andcotranslationally acquired a membrane-spanning conformation(shown in Fig. 1).

A second finding was that several AQP1 topogenic events arenot carried out with absolute fidelity in either expression sys-tem. Fusion proteins containing only the first three AQP1 TMsegments generated two different topological isoforms, a majortwo-spanning isoform in which TM3 C-terminal flanking resi-dues resided in the cytosol and a minor three-spanning isoformwith the TM3 C terminus in the ER lumen. Pulse-chase met-abolic labeling studies carried out in the presence and absenceof MG132 further demonstrated that these isoforms are differ-entially recognized and degraded by the proteasome-mediatedERAD pathway. Selective degradation of the two-spanning iso-form therefore resulted in a progressive increase in the per-centage of remaining polypeptides in the three-spanning (ma-ture) topology. This phenomenon was particularly apparent inHEK cells where very short pulse labeling times were requiredto confirm the presence of the predominant two-spanning iso-form. Its very short half-life of �25 min explains why theTurner group was unable to observe this species at steady stateand after a 1-h cyclohexamide chase (26). At these longer timepoints we too found that the three-spanning isoform predomi-nated. Although selective degradation also occurred in oocytes,the time scale of degradation (T1⁄2 � 15 h) was much slowerrelative to the time course of the experiments. Thus the two-spanning isoform was easily observed. Importantly, our resultsare completely compatible with data reported by the Turnergroup; however, we have reached different conclusions. Theprevious discrepancies in AQP1 biogenesis are primarily due tothe relative rates of protein synthesis and degradation in oo-cytes versus mammalian cells rather than fundamentally dif-ferent behavior of ER translocation machinery.

The findings that truncated fusion proteins can give rise tomultiple topological isoforms has several important implicationsfor our understanding of how polytopic protein topology is co-translationally established by the ribosome translocon complex.Given the rate of protein synthesis (�5 amino acids/s) in mam-malian cells (48) and the length of the ribosome exit tunnel (100Å) (49), TM segments and short connecting loops would normallyreside within the ribosome only for a few seconds before enteringthe translocon. Cotranslational topogenesis of small polytopicproteins such as AQP1 therefore requires that the transloconrapidly recognize topogenic determinants and dynamically directregions of peptide into their proper cellular compartment (50). Inthe case of AQP1 fusion proteins, the C-terminal reporter should

FIG. 6. MG132 blocks selective degradation of nonglycosylatedisoforms. A and C, EGFP-� subunit chimeras truncations 120 and 139(see Fig. 4) were expressed in HEK-293 cells. Pulse-chase labeling wasperformed in the presence of 20 �M MG132 as described in “Materialsand Methods. ” Glycosylated (�g) and nonglycosylated (�g) proteins areindicated. B and D, percentage of protein in the three-spanning (glyco-sylated) topology was quantitated for experiments carried out with(dark gray) or without (light gray) MG132. The data show the averagesof three experiments � S.E.M.

FIG. 7. Full-length AQP1 is stable in mammalian cells. Full-length AQP1 was expressed in HEK-293 cells. Pulse-chase labeling wasperformed in the absence or presence of 20 �M MG132 as describedunder “Materials and Methods.” Proteins glycosylated (�g) or nongly-coslated (�g) at Asn42 are indicated.

Stability of AQP1 Biogenesis Intermediates 267

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

theoretically reflect the last triage decision of the translocon andfollow the peptide loop into its appropriate compartment. Thisassumption serves as the basis for a large number of studies inwhich C-terminal reporters have been used to determine proteintopology (5, 11, 13, 38, 51). Our findings that different reporterdomains and AQP1 truncation sites yield similar topological out-comes in very different cell types support the argument thatchoice of the translocation pathway (either cytosolic or lumenal)is strongly determined by TM segments present within the ribo-some translocon complex. At the same time, however, certaintruncations appear to provide the reporter access to both the ERlumen and cytosol (albeit at different efficiencies) as evidenced bythe different topologic isoforms generated. This indicates that thetranslocation pathway is not gated absolutely in one direction oranother at all time points during AQP1 synthesis. Rather, thetranslocon appears to provide a certain degree of ambiguity inwhich TM segments and their flanking residues may transientlysample multiple topological configurations as the nascent chainextends from the ribosome. This finding is consistent with recentstudies demonstrating that N-terminal signal sequences can reg-ulate cytosolic accessibility of translocating proteins (and hencetranslocation efficiency) in a graded manner (52–54).

The substitution of large reporters such as the prolactindomain or the �-subunit in lieu of the native TM segmentslikely forces the translocon to make a decision as to where thepolypeptide should be directed based on the limited topogenicinformation available up to the particular truncation site. Wedo not yet know how flexible the environment is within theribosome-translocon complex nor how long a TM segment maybe allowed to sample different topological spaces. However, inthe case of AQP1, the relative proportion of different isoformsmay be viewed as reflecting relative accessibility of the reporterto ER and cytosolic compartments. Such flexibility in translo-con function might also facilitate proper reorientation of TM3at later stages of synthesis as more folding information isprovided. This is precisely what we observed using smallepitope tags inserted into the TM2–3 and TM3–4 peptide loops.TM3 topology gradually changed into its mature orientation asadditional TM segments and C-terminal folding informationwere synthesized (12, 23). Similar results have been reportedfor other polytopic proteins in which TM segment reorientationmay take place within the translocon (16).

One question that arises from these studies is why do differ-ent AQP1 topological isoforms exhibit such markedly differentstabilities. Because exposed hydrophobic patches can act assignals for ERAD (55), the two-spanning isoform could be de-graded because TM2 is exposed to the ER lumen (see Fig. 1A).In contrast, topology of the three-spanning isoform would moreclosely resemble the mature protein. The ERAD machinerycould therefore provide an efficient mechanism for removingproteins that exhibit abnormal topology in the ER membrane.Alternatively, stability may also be influenced by the reporter.Both reporters used here are normally expressed in ER lumen,and localization to the cytosol might therefore disrupt specificfolding requirements such as disulfide bond formation (prolac-tin) or glycosylation (�-subunit). Although we do not know theextent to which this occurs, the choice of a seemingly “inert”reporter may have significant implications when topologicstudies are carried out in intact cells with active ERAD path-ways and particularly when the reporter resides in both ERand cytosolic compartments. Our results demonstrate the im-portance of considering protein stability when performing to-pology studies with truncated intermediates because failure toexamine topology at early time points can lead to erroneousconclusions when multiple isoforms are generated.

This study provides the first direct comparison between the

early biogenesis mechanisms of oocytes and mammalian cellsusing a unique membrane protein substrate. Because oocytesprovide a useful and convenient system for heterologous ex-pression of many biologically important proteins (56), it isreassuring that early biogenesis events concur with those inmammalian cells. Indeed, when differences in protein stabilityare taken into account, the translocation efficiency and relativeproportion of different isoforms for all three truncations exam-ined here were in remarkably good agreement. This observa-tion held even when different reporters, reporter readouts, andfusion sites were compared. In addition, the topology of allother regions of AQP1 was essentially identical (22, 23). Ourfindings are therefore consistent with the strong conservationof ER translocation machinery (57–59) and indicate that theunusual biogenesis mechanism utilized by AQP1 reflects a wellconserved pathway of protein folding.

In this regard, we previously established that in Xenopusoocytes, once AQP1 synthesis is completed and all six TMsegments are present, the protein acquires a stable conforma-tion because of reorientation of TMs2–4 and repositioning ofthe TM2–3 and TM3–4 connecting loops (23). Selective degra-dation is therefore observed only for truncated fusion proteinsthat lack C-terminal folding information and are thereforetrapped in an unstable topology. Importantly, we also foundthis to be the case in mammalian cells. In contrast to constructscontaining only three TMs where the major isoform was de-graded with a T1⁄2 of �25 min, full-length AQP1 protein wasremarkably stable. Yet our data show that TM3 was cotrans-lationally directed into the unstable Nexo/Ccyt topology in 70%of these polypeptides. Thus generation of stable AQP1 proteinin mammalian cells, as in oocytes, is not achieved by selectivedegradation of an unstable four-spanning intermediate. In-stead, synthesis of C-terminal residues is able to confer stabil-ity on a protein region that is initially inserted into the mem-brane in an incorrect conformation. Given that mammaliancells and oocytes synthesize and process AQP1 into functionalwater channels with six TM segments (20, 60, 61) and thatAQP1 TM segments are cotranslationally directed into thesame immature topology in both systems, it seems reasonableto speculate that mammalian cells also have the capacity tocomplete AQP1 folding by facilitating topological reorientationof TMs 2–4 in the ER.

We should point out, however, that the purpose of our studywas primarily to compare cotranslational translocation eventsand does not address the mechanism(s) by which AQP1 ulti-mately acquires its mature six-spanning topology. Specifically,our studies do not address the efficiency of late traffickingevents but rather they examine stability of protein during earlymaturation in the ER which is where the six-spanning topologyis initially achieved (23). In this regard, our results contrastwith previous findings by Leitch et al. (47) that showed thatAQP1 degradation can be decreased by MG132 (and hypertonicstress) over longer time intervals. Additional work is thereforerequired to determine conclusively if and how AQP1 reorienta-tion takes place in mammalian cells. The extent to which thisunusual AQP biogenesis pathway is utilized by other proteinsand the mechanism by which this process is carried out withinthe ribosome translocon complex also represent important fu-ture challenges for unraveling the complex process of mem-brane protein biogenesis.

Acknowledgments—We thank J. Eledge for excellent technical assist-ance, Dr. R. J. Turner for generously providing the plasmidpEGFP.AQP1T120.�, Dr. A. S. Verkman for supplying AQP1 antisera,and members of the Skach lab for helpful comments.

Stability of AQP1 Biogenesis Intermediates268

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

REFERENCES

1. Alder, N. N., and Johnson, A. E. (2004) J. Biol. Chem. 279, 22787–227902. Turner, R. J. (2003) J. Membr. Biol. 192, 149–1573. Sadlish, H., and Skach, W. (2004) J. Membr. Biol., in press4. Blobel, G. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 1496–15005. Rothman, R. E., Andrews, D. W., Calayag, M. C., and Lingappa, V. R. (1988)

J. Biol. Chem. 263, 10470–104806. Lipp, J., Flint, N., Haeuptle, M. T., and Dobberstein, B. (1989) J. Cell Biol. 109,

2013–20227. Wessels, H. P., and Spiess, M. (1988) Cell 55, 61–708. Moss, K., Helm, A., Lu, Y., Bragin, A., and Skach, W. R. (1998) Mol. Biol. Cell

9, 2681–26979. Skach, W. R., and Lingappa, V. R. (1994) Cancer Res. 54, 3202–3209

10. Carveth, K., Buck, T., Anthony, V., and Skach, W. R. (2002) J. Biol. Chem. 277,39507–39514

11. Lin, J., and Addison, R. (1995) J. Biol. Chem. 270, 6942–694812. Wilkinson, B. M., Critchley, A. J., and Stirling, C. J. (1996) J. Biol. Chem. 271,

25590–2559713. Xie, Y., Langhans-Rajasekaran, S. A., Bellovino, D., and Morimoto, T. (1996)

J. Biol. Chem. 271, 2563–257314. Beguin, P., Hasler, U., Beggah, A., Horisberger, J. D., and Geering, K. (1998)

J. Biol. Chem. 273, 24921–2493115. Beguin, P., Hasler, U., Staub, O., and Geering, K. (2000) Mol. Biol. Cell 11,

1657–167216. Goder, V., Bieri, C., and Spiess, M. (1999) J. Cell Biol. 147, 257–26617. Fujiyoshi, Y., Mitsuoka, K., de Groot, B. L., Philippsen, A., Grubmeuller, H.,

Agre, P., and Engel, A. (2002) Curr. Opin. Struct. Biol. 12, 509–51518. Agre, P., King, L. S., Yasui, M., Guggino, W. B., Ottersen, O. P., Fujiyoshi, Y.,

Engel, A., and Nielsen, S. (2002) J. Physiol. 542, 3–1619. Fu, D., Libson, A., Miercke, L. J., Weitzman, C., Nollert, P., Krucinski, J., and

Stroud, R. M. (2000) Science 290, 481–48620. Sui, H., Han, B. G., Lee, J. K., Walian, P., and Jap, B. K. (2001) Nature 414,

872–87821. Schnermann, J., Chou, C. L., Ma, T., Traynor, T., Knepper, M. A., and Verk-

man, A. S. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 9660–966422. Skach, W. R., Shi, L. B., Calayag, M. C., Frigeri, A., Lingappa, V. R., and

Verkman, A. S. (1994) J. Cell Biol. 125, 803–81523. Lu, Y., Turnbull, I. R., Bragin, A., Carveth, K., Verkman, A. S., and Skach,

W. R. (2000) Mol. Biol. Cell 11, 2973–298524. Foster, W., Helm, A., Turnbull, I., Gulati, H., Yang, B., Verkman, A. S., and

Skach, W. R. (2000) J. Biol. Chem. 275, 34157–3416525. Shi, L. B., Skach, W. R., Ma, T., and Verkman, A. S. (1995) Biochemistry 34,

8250–825626. Dohke, Y., and Turner, R. J. (2002) J. Biol. Chem. 277, 15215–1521927. Skach, W. R., and Verkman, A. S. (1995) Biophysical J. 68, A334 (abstr.)28. Tamarappoo, B. K., and Verkman, A. S. (1998) J. Clin. Investig. 101,

2257–226729. Sasaki, S., Fushimi, K., Saito, H., Saito, F., Uchida, S., Ishibashi, K., Kuwa-

hara, M., Ikeuchi, T., Inui, K., Nakajima, K., Watanabe, T., and Marumo, F.(1994) J. Clin. Investig. 93, 1250–1256

30. Mulders, S. M., Knoers, N. V., Van Lieburg, A. F., Monnens, L. A., Leumann,E., Weuhl, E., Schober, E., Rijss, J. P., Van Os, C. H., and Deen, P. M. (1997)

J. Am. Soc. Nephrol. 8, 242–24831. Deen, P. M., Croes, H., van Aubel, R. A., Ginsel, L. A., and van Os, C. H. (1995)

J. Clin. Investig. 95, 2291–229632. Deen, P. M., Verdijk, M. A., Knoers, N. V., Wieringa, B., Monnens, L. A., van

Os, C. H., and van Oost, B. A. (1994) Science 264, 92–9533. Laemmli, U. K. (1970) Nature 227, 680–68534. San Millan, J. L., Boyd, D., Dalbey, R., Wickner, W., and Beckwith, J. (1989)

J. Bacteriol. 171, 5536–554135. Boyd, D., Traxler, B., and Beckwith, J. (1993) J. Bacteriol. 175, 553–55636. Skach, W. R., Calayag, M. C., and Lingappa, V. R. (1993) J. Biol. Chem. 268,

6903–690837. Tector, M., and Hartl, F. U. (1999) EMBO J. 18, 6290–629838. Bayle, D., Weeks, D., and Sachs, G. (1997) J. Biol. Chem. 272, 19697–1970739. Ota, K., Sakaguchi, M., von Heijne, G., Hamasaki, N., and Mihara, K. (1998)

Mol. Cell 2, 495–50340. Hubbard, S. C., and Ivatt, R. J. (1981) Annu. Rev. Biochem. 50, 555–58341. Kaplan, H. A., Welply, J. K., and Lennarz, W. J. (1987) Biochim. Biophys. Acta

906, 161–17342. Hartmann, E., Rapoport, T. A., and Lodish, H. F. (1989) Proc. Natl. Acad. Sci.

U. S. A. 86, 5786–579043. Xiong, X., Bragin, A., Widdicombe, J. H., Cohn, J., and Skach, W. R. (1997)

J. Clin. Investig. 100, 1079–108844. Buck, T. M., Eledge, J., and Skach, W. R. (2004) Am. J. Physiol. 287,

C1291–C129945. Nilsson, I., and von Heijne, G. (2000) J. Biol. Chem. 275, 17338–1734346. Dohke, Y., Oh, Y. S., Ambudkar, I. S., and Turner, R. J. (2004) J. Biol. Chem.

279, 12242–1224847. Leitch, V., Agre, P., and King, L. S. (2001) Proc. Natl. Acad. Sci. U. S. A. 98,

2894–289848. Braakman, I., Hoover-Litty, H., Wagner, K. R., and Helenius, A. (1991) J. Cell

Biol. 114, 401–41149. Ban, N., Freeborn, B., Nissen, P., Penczek, P., Grassucci, R. A., Sweet, R.,

Frank, J., Moore, P. B., and Steitz, T. A. (1998) Cell 93, 1105–111550. Johnson, A. E., and van Waes, M. A. (1999) Annu. Rev. Cell Dev. Biol. 15,

799–84251. Manoil, C., Mekalanos, J. J., and Beckwith, J. (1990) J. Bacteriol. 172,

515–51852. Fons, R. D., Bogert, B. A., and Hegde, R. S. (2003) J. Cell Biol. 160, 529–53953. Kim, S. J., and Hegde, R. S. (2002) Mol. Biol. Cell 13, 3775–378654. Kim, S. J., Mitra, D., Salerno, J. R., and Hegde, R. S. (2002) Dev. Cell 2,

207–21755. McCracken, A. A., and Brodsky, J. L. (2003) Bioessays 25, 868–87756. Wagner, C. A., Friedrich, B., Setiawan, I., Lang, F., and Broer, S. (2000) Cell

Physiol. Biochem. 10, 1–1257. Matlack, K. E., Mothes, W., and Rapoport, T. A. (1998) Cell 92, 381–39058. Eichler, J., and Duong, F. (2004) Trends Biochem. Sci. 29, 221–22359. Van den Berg, B., Clemons, W. M., Jr., Collinson, I., Modis, Y., Hartmann, E.,

Harrison, S. C., and Rapoport, T. A. (2004) Nature 427, 36–4460. Katsura, T., Verbavatz, J. M., Farinas, J., Ma, T., Ausiello, D. A., Verkman,

A. S., and Brown, D. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 7212–721661. Toriano, R., Ford, P., Rivarola, V., Tamarappoo, B. K., Verkman, A. S., and

Parisi, M. (1998) J. Membr. Biol. 161, 141–149

Stability of AQP1 Biogenesis Intermediates 269

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from

Teresa M. Buck and William R. SkachAquaporin-1 Topological Maturation

Differential Stability of Biogenesis Intermediates Reveals a Common Pathway for

doi: 10.1074/jbc.M409920200 originally published online October 28, 20042005, 280:261-269.J. Biol. Chem. 

  10.1074/jbc.M409920200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/280/1/261.full.html#ref-list-1

This article cites 60 references, 35 of which can be accessed free at

by guest on February 23, 2016http://w

ww

.jbc.org/D

ownloaded from